From: Patrick Palka <patrick@parcs.ath.cx>
Date: Thu, 18 Jun 2015 22:25:45 GMT
Subject: Re: [PATCH] Improve contrib/diff-highlight to highlight unevenly-sized hunks
Message-ID: <CA+C-WL97BeK3JmSPdyOdg1j=QOFTht_R=hUoAA30Vq6Nqx-wdQ@mail.gmail.com>
In-Reply-To: <20150618212356.GA20271@peff.net>

On Thu, Jun 18, 2015 at 5:23 PM, Jeff King <peff@peff.net> wrote:
> On Thu, Jun 18, 2015 at 04:45:05PM -0400, Jeff King wrote:
>
>> Still, I think this is probably a minority case, and it may be
>> outweighed by the improvements. The "real" solution is to consider the
>> hunk as a whole and do an LCS diff on it, which would show that yes,
>> it's worth highlighting both of those spots, as they are a small
>> percentage of the total hunk.
>
> I've been meaning to play with this for years, so I took the opportunity
> to spend a little time on it. :)

Cool!

>
> Below is a (slightly hacky) patch I came up with. It seems to work, and
> produces really great output in some cases. For instance, in 99a2cfb, it
> produces (I put highlighted bits in angle brackets):
>
>   -               <hash>cpy(peeled, <sha1>);
>   +               <oid>cpy(<&>peeled, <oid>);
>
> It also produces nonsense like:
>
>   -       <un>s<ign>ed <char >peeled<[20]>;
>   +       s<truct obj>e<ct_i>d peeled;

That's not even so bad.  The diff of the change itself is... interesting.

>
> but I think that is simply because my splitting function is terrible (it
> splits each byte, whereas we'd probably want to use whitespace and
> punctuation, or something content-specific).

I hope you can polish this.  It definitely has potential.